Yes, We Are At A Tipping Point: ChatGPT Is Just The Beginning Of How AI Will Soon Change Everything

hi everyone I'm Diane Brady I'm here with my colleague Kendrick Tai who is a senior reporter Silicon Valley Kenrick welcome we're talking I guess chat gbt and uh all things generative AI this week based on let's talk a little bit about the the daily cover that you and Alex Conrad did starting with um are we at a Tipping Point because I feel like we've had so much news around this including what you've covered it just seems like it's escalating yeah absolutely well thanks for having me Diane um I think you know Alex and I we spoke to 60 plus people in the AI space Founders investors researchers and I guess the backdrop with AI is you know there have been so many hype Cycles in the past right like uh IBM's deep blue beating Gary kasparova chess back in uh 1990 age yeah Generations ago and there's just been like multiple hype Cycles where it's like you know this AI is going to you know take over the world or it's going to like change the way you work and uh I think it hasn't really manifested in a major way uh in the business side of things um it's caused a lot of investors to lose a lot of money because they made these bets that were speculative and it really seemed like it was taking off but then you know the business case didn't really take off um but I think what we're seeing now is that we are at a new Tipping Point um and things are really starting to take off I think uh what's really new about you know this latest AI hype cycle that's uh different and a lot more immediate than the past type Cycles is that you know it's like kind of the magic is right in the hands of the consumers you know like they can just type in a sentence into a tool like chat gbt or dolly or stable diffusion and then they can get you know whole paragraphs answer or like a whole generation they can do your essay for you and pass the Wharton test and everything else you mentioned other companies so much of the attention's been on chat GPT and I know of course you know um you guys spoke to Sam Altman and open Ai and such it's sucking the air out of the room is it markedly better than everyone else or are there a lot of players out there yeah it's uh it's looking very uh very much too early to tell right now I think what makes open AI stand out from the crowd right now is um a they're just you know the highest valued company out of the startups in the space you know their ad thank you Sachin Adella yeah it's a it's a big investment reportedly uh so you know they have that big lead they have you know a lot of that um from a metric standpoint they have you know a lot of computing power from Microsoft and that's really kind of made them stand out uh in terms of getting a lot of these tools out there but from a technological standpoint I think it's still too early to tell like who's really going to be that winner there um we spoke for example to Alex Wang at scale who provides a lot of the infrastructure tools the picks and shovels to help these AI model companies to you know make their models and he was talking to me about how you know openai put out this uh tool chat GPT um this month or I guess this would be in January um another company called anthropic which was founded by a couple of former open AI researchers they put out their own uh kind of chatbot tool called Claude and how does Claude compare to chat GPT yeah basically what uh scale I think did like a sort of analysis of like the two and like inputting different use cases like who's better at coding or who's better at answering like a mathematical question and what they ended up finding out was like it was kind of a dead heat you know it was uh they're both good at their own things but neither of them was markedly better than the other and what's really interesting there is you know open AI has spent significantly more money trading chat GPT than anthropic did Claude so um basically I think the conclusion there is it's just way too early to tell like who is actually going to win this race um the kind of differentiation between the models is relatively similar right now because a lot of these companies are working with you know the same buckets of data just like all the data and that's that's an interesting point because if you step back and look at say the race for you know HIV cure any sort of you know that that the technology is developing in a pool and everybody essentially has access to it so um it's not like there's a Eureka moment this really is the culmination of years and years of research and work that's been done into like talk a little bit about some of the other potential winners in the space you spoke to 60 uh incredible number of people who are working here any other companies and names that emerged from your reporting sure yeah um well I think the big thing to consider with this whole AI race is that uh kind of the biggest benefactors right now are not necessarily these new companies or not even like a I guess open AI isn't necessarily an upstart anymore but they are you know still one of the newer players in the arena so far I think the biggest benefactors have been just a lot of the cloud providers the uh you know big Tech folks um Google and Microsoft and Amazon that are providing a lot of the computing power and I think we're starting to see now that uh these companies are you know kind of getting a horse into the race in a partnering with different companies in the space and so that provides a big advantage to certain companies you know open AI with Microsoft we were reporting on stability AI who has a very deep partnership with Amazon um where they have some kind of arrangement where Amazon provided stability with more than 4 000 Nvidia gpus like these AI chips that put together kind of created one of the world's biggest supercomputers for stability to run their models on so um this really popular image generation tool called stable diffusion um was uh really made possible and Powerful by that partnership um besides that I think we are seeing just a lot of new companies starting to emerge um one of the big kind of uh offshoot points um was this paper that a bunch of Google researchers put out in 2017 called attention is all you need and it basically uh invented this new paradigm um for it's called the Transformer that uh is kind of a new machine learning approach that has kind of made this whole generative boom kind of possible um and I guess it was just a really great technological breakthrough and what we've seen now is like six of the eight researchers on that paper have left to start their own AI companies and a lot of those are some of the most promising companies right now so you have companies like cohere and Adept and character AI uh but what space are they working and when you talk about those companies um I mean are they are they basically um working on specific use cases for gender of AI is it sort of helping to develop the technology further or where would you say the general neighborhood is yeah it's a good question um I think for some of those com so I think there's two different categories and for some of the companies I was just mentioning open AI stability cohere Adept uh they're creating kind of the models the kind of base layers um that uh you know that these new AI breakthroughs are coming from so one example being um catch EBT the underlying basis is you know something called GPT which is a um sort of text generation model and then we're talking about uh stable diffusion earlier which is a sort of text to image model um and then there are these kind of other like models that are um being developed and that's kind of like the uh foundational layer for all this AI but then what we're also seeing is all these applications that are being built on top of the AI models so I think something popular example might be like lenza the uh kind of selfie app that uh you just uh upload a selfie of yourself and then it kind of generates all these like really cool outlandish uh selfies of you um on a more sophisticated basis than we can already do with like dog ears and tongues and such obviously yeah absolutely it'll like make you look like a uh astronaut or like a medieval warrior or something like that check that out my new my new profile let me ask you know one of the things what happened let's say with Cloud was it did really sort of provide um a foundation for this whole zero marginal cost economy where you saw this explosion of new businesses that were able to be created I think I'm still curious as to whether this is really reinforcing the power of the Titans or if we're seeing another opportunity here where we it will unleash new creativity new startups probably a bit of both but what sense are you getting yeah I'd say uh it's a big question that I was trying to get a sense of from talking to all these people and it really does seem like the jury is still out but uh I think to some extent like this does most likely uh bolster the power of the Titans um kind of going back to for a while and like the uh Computing cost of all this you know the these Cloud providers tend to just benefit a lot from um you know powering these AI uh this AI in the first place um I think another thing that makes this AI Paradigm Shift different from certain other Paradigm shifts is that uh a lot of this technology is very easy to integrate like even with something like lenza that's a totally new app that's built on top of um stable diffusion um the image generation model that we were talking about earlier and that was you know pretty relatively easy to kind of just like integrate that into an app and kind of just put this app out there and they're making like at their Peak according to um this company sensor tower that uh tracks like app store data they're making like 2.5 million dollars at a single day at their Peak just wow quickly uploading uh it's like Angry Birds territory right yeah um I think that kind of speaks to the point to your question of like the big Tech um for them like a it's like even easier to be able to kind of integrate this technology and then you know with companies like and use it as a service offering in essence like as part of an Enterprise package here now we've got this you can do a b and c yeah potentially I mean um this kind of feels very new in that uh this technology is kind of blowing up in these last couple months but uh Google has had like a AI arm for 15 plus years that you know employs thousands of people and uh they have this technology kind of and a lot of it has been behind the scenes um we're starting to see it a little bit like if you use Gmail there's like smart composed where it'll like automatically suggest like the next here's what you really want to say to your mother right that kind of thing well well let me you know one of the things that I think um in terms of the Zeitgeist there is the Euphoria fear Nexus and it feels like at this point the economy is down so there's job loss and there's concern about what the job gains going to look like now that we've got these new tools there's already concern about deep fakes and Trust on the internet and now we've got these tools in general um what is what is the reaction around the use cases I mean we're already seeing AI generated you know Seinfeld like comedies and as I mentioned programs that are passing tests that Wharton it does inspire a certain degree of fear as to where the human is in the equation what are you sensing in terms of what the repercussions are of that yeah that's a that's a very interesting existential existed I think uh I mean to kind of tie a bow out of your previous question too like we're seeing and what we're probably going to see like first before anything gets you know to Bonkers is uh just the application of these AI uh breakthroughs to business tools so I think we'll see like more and more like for example Microsoft integrating into word and Outlook or um chat Bots chat Bots uh the different tools that Google has and that you know meta can apply this AI to Facebook um and we'll probably see it more and more in uh business use cases like that different productivity tools and things like that but um I think what you're touching on as the potential existential threat is uh very much something that uh people are starting to think about and be concerned about and uh you start to be thoughtful about Universal basic income comes back into the conversation perhaps as we look at jobs coming back um any thoughts from the regulatory front I mean is this when we talk about a Tipping Point um you know it does that that implies that we are in a sort of new playing field and a new level of Interest what what are we tipping into yeah um a lot of a lot of unanswered questions I think I mean even to the existential Threat all right there's this uh term called artificial general intelligence uh AGI which kind of um stands for uh this sort of hypothetical future AI that is sentient and like self-conscious and um kind of gets to all the Sci-Fi Promises of like what AI could be right Comfort your dad or start a war it depends yeah um and I think one of the really interesting takeaways from all the reporting that Alex and I did uh Alex was talking to Sam Altman at openai who one of their company missions or their primary mission is to kind of build towards this AGI and what Altman was telling us was um it's quite possible that when we get there to full AGI like we won't even recognize that we've already gotten there um so there are a lot of like kind of interesting repercussions or potential like exciting things that happen out of that um up and was saying like that's the type of thing that could potentially you know break capitalism it could be something that like you said like reintroduces this whole Universal basic income thing to the equation um I think before we get that there are a lot of like just legal questions to be figured out you know like on a some of them are kind of more more on a very like basic regulatory front like what is the um what are IP rules here what are the copyright rules for this whole new paradigm shift you know like open Ai and Microsoft have been sued for purportedly kind of by reading um quote unquote uh like programmers code and similarly like stability has a lawsuit from getting images for supposedly or allegedly kind of stealing all these images off of Getty servers and kind of using it we have to teach the program morals perhaps you know one of the things that has come up of course is just the whole nature of algorithmic bias you know I I and and some people have experimented with how they describe say Biden versus Trump or you know write me a poem and and um how much is that from just to kind of you know tie this up from where you said as somebody who's covering this um what are the questions that you have in your mind both and the concerns and the opportunities let's start with the questions you have coming out of this exercise of essentially doing a deep dive on something we're just hearing about you know bits and pieces every week yeah totally um I think it would be two things um one of them is like how do we resolve these issues of um bias inaccuracy which are still you know very much still there I think they were they've been kind of a hot debate Point as far as this AI technology goes for for years but uh just because we've reached this kind of Tipping Point uh from a business standpoint doesn't mean those issues have been resolved yet and um from talking to some of the folks uh for for this story you know we found like when chat GPT writes a job description there are still elements of uh bias in there and uh interesting takeaway from a conversation with uh Kieran Snyder who is a CEO of a company called uh textio I think which kind of is a recruiting startup um but she inputted like some of these uh job descriptions into chat GPT and some of the most compelling like uh job recruiting descriptions were the ones that had the most bias so you know it's still very much an issue there and it's still very much an issue also in terms of like what is accurate right like if you type something into uh chat gbt um it'll very confidently like give you an answer but the answer might not necessarily be the correct answer so um people in the AI space call that kind of hallucination and there's still a lot of this like AI hallucinating giving you these like answers that if you take at face value they could like kind of screw up your business or kind of screw up your uh homework fire everybody right well we uh We've certainly seen examples of that without a machine operated tool I mean um yeah sorry to interrupt but uh I think the second thing is just like the more uh long-term existential question that we were kind of touching on earlier um I think kind of as far as regulations and as far as like who is making the decisions go uh we're just barely at the uh at the start of that um we could see like kind of a 2010s type of you know all these Congressional hearings with these social media companies um we saw how that went yeah and talking to some of these researchers in the AI space uh it seems like that could very much be a similar thing that will happen with certain AI companies where they'll need to answer to different regulations in different places uh we spoke to a researcher at Harvard uh Aviv ovadia who said you know kind of we need policy making at the speed of you know this technology and um that's because you know this could have really big um impacts on just the way that you know basic core tenants of our democracy like free speech and you know dissemination of Truth or information like how that stuff is affected like at a very onerous hypothetical like these things could be used to generate like fake videos of you know a violent Riot happening or something like that right which which is which is interesting I mean one of the things um you know two other points I think are worth touching on right now one is that you know as we're saying this there is a alleged you know Chinese uh a balloon that's you know it's buying us we've seen conversations about Tick Tock and Congress you know obviously there's a lot of innovation happening right now in China you've got the regulatory environment in Europe let's step outside the U.S I know this wasn't the focus of your story but how much is what are we ahead of the curve or are we really sort of operating in parallel especially with China where there's so much Innovation we haven't been talking about yeah it's a it's a good question um to be determined perhaps because again it's not the most transparent of places when it comes to that I do think a lot of it is to be determined but uh China is very much a prolific um threat or competitor however you want to perceive that um as far as this race goes um I think there have been statistics out about how China has now lapped the US in terms of the uh amount of AI research papers that are coming out of China versus coming out of the United States and um you know there's very much a kind of wild garden situation uh where China is developing its own set of AI Technologies and the US is developing our own set of AI Technologies and so it's still kind of TBD like where that's going to go but uh I think there's very much a big question there that is like causing um US government for example to kind of start looking at Silicon Valley deeper and deeper you know whereas like China very much they're kind of figuring out their own strategies like this AI chips act for example that Biden uh put forth that kind of uh could be a damper for China's efforts you know not being able to acquire so many Nvidia chips anymore but uh I think you know they have their own kind of plays that are uh still coming and one other question Kenrick which is um you know there's a certain kind of orwellian Fear Factor tone I know that has been certainly in the headlines um but in Silicon Valley I also think there must be a lot of conversation around the opportunity the excitement especially in Realms like Health Care Etc what what is the general tone you're hearing and where do you think are the particular pockets of excitement that we should watch in terms of how this plays out and how it can transform certain industries yeah totally I think um I think it's the the overall sentiment in Silicon Valley around this technology is like uh very excited one um I mean I feel like all the conversations that I was having at the end of uh 2022 with investors was just like he's you know onerous times as far as the market goes and then now we're looking at what's happening in AI investment and it's as if uh it's as if this uh this Market were not the way it was you know it's as if like everything was still a year a new day right yeah but I think like that kind of the reason for that is because um kind of this AI has gotten to that Tipping Point where um people are starting to imagine all sorts of things that uh that could completely change you know the way that we work or the way that we live you know and I think the most exciting companies to come out of this technology probably have not been founded yet and you know are probably in use cases that uh we haven't really seen the AI touch yet you know like I think the most exciting places are like healthcare or education you know and just like thinking about um a kind of basic like next step example right like can chat GPT get to the point where it can be a personalized one-on-one therapist or personalized one-on-one tutor something like that but um you know we're seeing like a lot of developments in drug Discovery and like figuring out how to cure different uh diseases and um I think there are a lot of promising ways that this technology is going to just really supercharge um hopefully benefits for Humanity uh on the other hand guardrails are necessary to make sure exciting days to come this is no doubt the year of generative Ai and look forward to continuing the conversation thanks for joining us Kendrick thanks so much for having me Diane

Channel: Forbes

Views: 149,472

Keywords: Forbes, Forbes Media, Forbes Magazine, Forbes Digital, Business, Finance, Entrepreneurship, Technology, Investing, Personal Finance

Id: tcnvSQZ6rFo

Length: 26min 32sec (1592 seconds)

Published: Mon Feb 06 2023